Thermomechanical Stability

The TBC system is expected to survive the thermal shock of rapid cycle heating and cooling, as experienced in gas turbine or diesel engines, without spallation. Spallation is the result of the thermal stress pattern created when the thermal cycle repeatedly exceeds the stress for crack growth within the zirconia layer. This results in a time-dependent extension of pre-existing cracks within the plane of the coating until critical link-up and spallation. In many tests, the location of the spall line is in the zirconia layer, but near the bond coat interface. In the finite element modeling studies of Chang et al. (Ref 1), the near-interface area in the zirconia layer was shown to be the high-stress zone in simulated thermal cycling. In the thermal spray process, the crack nuclei are probably the interfaces between splats in the coating.

Two strategies to improve the resistance to in-plane crack growth in thermally sprayed zirconias are to use a low-density coating, with the pores acting to blunt growing cracks, or to increase the cohesive strength between splats, which requires high-density coating conditions. In the first case, 12 to 15% porosity is useful, and in the latter case, it is critical to generate macrocracks vertical to the plane of the coating, spaced about 0.2 to 1 mm apart, to relieve short-range stress (Ref 2). In physical vapor deposition coating, the zirconia deposition conditions are adjusted to grow columnar grains, which similarly reduce the in-plane modulus and limit the accumulation of coating stress during a thermal cycle.

Thermal Fatigue Testing. Laboratory testing has been used, particularly in the coating development stage, to cycle thermally a TBC specimen followed by post-test microscopic examination for spallation-type cracks. The thermal cycle for evaluation of thermal shock resistance employs rapid heating and cooling rates, using direct impingement flames or heating jets on the oxide face of the specimen, with little hold time at the maximum temperature. This test principally challenges the oxide layer, because the bond coat remains at relatively low temperatures due to the insulating nature of the zirconia layer and the short time at high temperatures.

Figure 1 illustrates a cycle that simulates the thermal shock of first-stage gas turbine outer airseals with a zirconia layer 1.27 mm (0.050 in.) thick. Detail of the flame impingement on the button sample is shown in Fig. 2. The test sample is a 25 mm (1 in.) diameter button, 3.2 mm (0.125 in.) thick, made from the substrate alloy of interest. After the button is coated with the TBC system, the edges are ground and polished to produce a square edge to allow examination of the coating layers.

1210

1020

1210

1020

^ 2550 °F

JpH ™

i Heat

-

\ "■■■ blast \ coo!

\ cool J \ y I

\ -

i i i V

1500

1000

2000

1500

1000

at a

20 40 60 80 100 120 140 Time, s

Fig. 1 Schematic illustration of oxide surface temperature cycle used for thermal fatigue testing of ZrO2-coated test buttons 1.27 mm (0.050 in.) thick, simulating a first-stage gas turbine outer airseal application

Fig. 1 Schematic illustration of oxide surface temperature cycle used for thermal fatigue testing of ZrO2-coated test buttons 1.27 mm (0.050 in.) thick, simulating a first-stage gas turbine outer airseal application

Fig. 2 Flame impingement on the button sample in a thermal fatigue test rig

The test rig consists of an oxygen-propane or propylene burner, a fixture arm to hold the button sample, a stepping motor system to advance quickly between stations and to hold for prescribed times in the heating and cooling positions, and a means to record at least the front oxide face temperature during the heating cycle. Two-color infrared optical pyrometers are widely used for the latter task, with the output connected to a chart recorder or computer. One word of caution concerning the temperature measured. If the zirconia coating is not an ideal gray body, that is, the emissivity is not the same at the two wavelengths used by the pyrometer, the reading may be in error. This issue is under study at several test facilities, and it could affect burner rig tests as well. If the samples being tested are all the same material, such as plasma-sprayed 8% YSZ coatings of common thickness, the measured temperatures should be relatively correct.

In one test configuration, the heating cycle is 20 s, followed by a 20 s air blast and two 20 s periods of natural convection cooling. The burner-to-specimen standoff, burner size, and gas flows are set to heat the oxide face of a standard specimen rapidly to 1400 °C (2550 °F) in the first 20 s. The air blast then drops the front face temperature to about 815 °C (1500 °F), and it finally reaches about 454 °C (850 °F) after 40 s of natural cooling. After the cycle is repeated 2000 times, the edge of the coated button is examined at 10* to 30* for evidence of separation-type cracking in the zirconia layer. The specimen edge should also be inspected before the test, with few to no starting cracks expected.

The post-test cracking of good TBC systems having zirconia thicknesses of about 1.27 mm (0.050 in.) should be less than 15% of the circumference, and typically much less. If a tested specimen is mounted in cross section and serially polished, it will be seen that the crack is indeed at the edge of the coating and extends inward toward the center of the button. The cracks will continue to extend around the circumference and grow inward until eventual spallation, for the case of a TBC with poor thermal fatigue resistance. In good TBC systems, additional sets of 2000 cycles will show little crack growth, if any, and typically at a lower rate than in the first test period.

It has been found that the edge-cracking rate is a function of the zirconia coating density (Ref 2), as shown in Fig. 3. In addition, use of the conventional tensile bond-cap test (ASTM C 633-79) has shown that the higher-density coatings have increasing cohesive strength within the zirconia (Ref 2). The result of Fig. 3 must be further qualified. The coatings of density above about 90% of theoretical density (below 10% porosity) also had intentional long macrocracks throughout the coating running perpendicular to the coating plane. Without these macrocracks, the higher-density zirconia coatings will spall quickly, perhaps even on the first thermal cycle.

CD CL

Cj CD

6 5

oo

-

, 7

3

-

Ou

0

-

6

-

O

-

1A

21

-

i i

1 1

©

■ (3D

flß.O 87SS,CI 39,0 90.0 91.0 Theoretical density, %

92.0

Fig. 3 Thermal fatigue edge-cracking rate for thick ZrO2 coatings as a function of coating density. Thermal fatigue test with peak temperature of 1400 °C (2550 °F) and 1.1 to 1.3 mm (0.4 to 0.5 in.) thick oxide. Specimen numbers allow correlation to Fig. 4. Source: Ref 2

If the back metal side of the button specimen also has temperature instrumentation, the differential front-to-back temperature drop, AT, can be measured at the peak of the heating cycle. Figure 4 shows that AT is also related to coating density, with lower-density coatings having greater thermal insulation.

BOO 750 700 650 BOO 550 500 450

4 _

-

3

O

-

-

IN*

IB

-

Vo

\ 2

-

-

1

l l l

1200

1100

1000

86.0 37.0 38.0 89.0 90.0 91.0 92.0 Theoretical density %

Fig. 4 Temperature drop across thick ZrO2 coatings on an IN 718 substrate (thickness, 3.2 mm) as a function of ZrO2 coating density. Thermal fatigue test with peak temperature of 1400 °C (2550 °F) and 1.1 to 1.3 mm (0.4 to 0.5 in.) thick oxide. Specimen numbers allow correlation to Fig. 3. Source: Ref 2

The same rapid heating test rig can be used to evaluate thin TBCs. On aircraft turbine components such as first-stage blades or vanes, the zirconia thickness required may be from 0.13 to 0.30 mm (0.005 to 0.012 in.), and on combustion chambers, 0.30 to 0.38 mm (0.012 to 0.015 in.). If the front face oxide peak temperature is set at 1400 °C (2550 °F) on a standard YSZ sample of 90 to 91% theoretical density and 1.27 mm (0.050 in.) thickness, specimens with thinner zirconia layers will reach lower temperatures in the same test cycle, while those with thicker layers will reach higher temperatures. Figure 5 shows this dependence on oxide layer thickness when other conditions are constant. For this reason, the burner parameters should be set using a standard sample to ensure a constant heat flux for all tests. One way to do this is to have a rotary carousel of samples, including one standard sample, that provides a check on the test conditions every time it cycles through the heating station. The parameter AT will also increase with zirconia thickness, as one would expect. Furthermore, the edge-cracking rate will depend on coating thickness for constant coating density and structure. For example, a dense macrocracked TBC specimen 0.30 to 0.38 mm (0.012 to 0.015 in.) thick can easily sustain 20,000 or more of the above cycles without edge cracking.

Fig. 5 Peak temperature of zirconia surface for ZrO2 coatings as a function of coating thickness, in standard thermal cycle test of Fig. 1 (20 s cycle to 1400 °C, or 2550 °F). All samples were run at the same relative heat flux (1330 °C, or 2423 °F, on a nominal coating 1270 pm, or 50 mils, thick).

Fig. 5 Peak temperature of zirconia surface for ZrO2 coatings as a function of coating thickness, in standard thermal cycle test of Fig. 1 (20 s cycle to 1400 °C, or 2550 °F). All samples were run at the same relative heat flux (1330 °C, or 2423 °F, on a nominal coating 1270 pm, or 50 mils, thick).

A similar thermal fatigue test cycle can be established for testing TBCs for the diesel environment. In this case, the heat flux across the TBC is expected to be greater than in gas turbines, but the peak temperature is expected to be lower. In one test configuration, the burners were moved closer to the oxide face of the coated button, the heating cycle was reduced to 8 s, and the peak oxide temperature was held to 982 °C (1800 °F).

Another testing approach, used by aircraft gas turbine manufacturers and the National Aeronautics and Space Administration, is to coat solid 1.2 cm (0.5 in.) diameter burner bars and expose a rapidly rotating carousel of bars to a high-velocity fuel-oxygen burner. This test has the advantage of allowing use of commercial engine fuel, with controlled doping of known impurities such as sulfur or vanadium if desired. Substrate-side air cooling is sometimes introduced by using hollow bars, to make the test a better simulation of actual coated hardware conditions where back-side cooling is typical.

Although the button tests establish the capability of the basic coating candidate, the button and burner bar coating conditions are usually ideal. Only on real components will practical issues such as the effect of component geometry on the coating structure be faced. One further stage of testing might be considered: putting actual coated components in large thermal cycling chambers. Even so, the laboratory screening tests described above should be followed by actual engine tests of components with TBCs.

References cited in this section

1. G.C. Chang, W. Phucharoen, and R.A. Miller, Finite Element Thermal Stress Solutions for Thermal Barrier Coatings, Surf. Coat. Technol, Vol 32, 1987, p 305-325

2. T.A. Taylor, D.L. Appleby, A.E. Weatherill, and J. Griffiths, Plasma-Sprayed Yttria-Stabilized Zirconia Coatings: Structure-Property Relationships, Surf. Coat. Technol., Vol 43/44, 1990, p 470-480

0 0

Post a comment